We examined the incidence of zoster in 346 children with underlying acute lymphoblastic leukemia who were immunized with live attenuated varicella vaccine while in remission. We also compared a subset of 84 of these children with a matched group of 84 children with leukemia who had had natural infection with varicella. Of the 346 vaccinated children, 5 (1.45 percent) became infected with zoster after 10,878 months of observation, for an incidence of 0.552 case per 100 person-years. Among the matched pairs of subjects, zoster occurred in 3 (3.6 percent) of the 84 vaccinated subjects during 2936 months of observation--an incidence of 1.23 cases per 100 person-years--and in 11 (13.1 percent) of the subjects with natural infection during 4245 months--an incidence of 3.11 cases of zoster per 100 person-years. Although the incidence of zoster was more than twice as high in the control children as in the vaccinated children (3.11 vs. 1.23 cases per 100 person-years), a Kaplan-Meier product-limit analysis revealed no significant differences in incidence between the two groups. Children from both groups in whom leukemia recurred were more likely to contract zoster than those who did not have a recurrence (7 of 35 vs. 7 of 133, P less than 0.025). Zoster was not a marker for impending relapse. No case of zoster was severe or disseminated. We conclude that the incidence of zoster following immunization with live attenuated varicella vaccine is no greater than that following natural varicella infection.